Beijing plans curbs on overseas access to China's top AI models: Report

Chinese authorities have reportedly held meetings with leading technology companies over the past month to discuss potentially restricting overseas access to China's most advanced AI models, according to people familiar with the matter who spoke to Reuters.

The discussions follow a series of steps taken by Beijing to keep homegrown AI technology within the country and underscore that China may soon begin treating cutting-edge AI as a critical national asset requiring export controls, similar to those imposed by the United States.

The discussions reportedly included representatives from several of China's leading technology companies, including Alibaba, ByteDance, and AI startup Z.ai, according to Reuters. Chinese AI models have made significant inroads into global markets, thanks to their lower costs and rapidly improving capabilities since the emergence of DeepSeek's R1 model last year. Any decision to restrict overseas access to these models could have far-reaching implications for the global AI industry.

More from Tech

Toughening penalties for AI exports

The meetings, led by China's Ministry of Commerce, reportedly focused on imposing tighter controls on the country's most advanced AI models, including both closed-source and open-source systems, according to two people familiar with the discussions. Officials also discussed introducing new measures that could restrict who is allowed to fund domestic AI startups, as Beijing seeks greater control over its rapidly growing AI ecosystem.

All three companies that attended the meetings have developed widely used AI models. Alibaba's Qwen and ByteDance's Doubao are among China's most popular AI models, while Z.ai recently attracted significant attention in Silicon Valley after unveiling its GLM-5.2 model, which reportedly delivers performance comparable to some leading US AI systems at a fraction of the cost.

National security concerns

The reported discussions come amid growing global concerns over the strategic implications of advanced AI. In the past, US President Donald Trump has expressed concerns that cutting-edge American AI models could be exploited by military and intelligence agencies in China, Russia, and other countries.

In June, the US government ordered that foreign nationals be denied access to Anthropic's most advanced Fable and Mythos models. Export restrictions on Fable were later eased after Anthropic introduced additional safeguards.

As part of its broader efforts to protect homegrown AI, Beijing has introduced a series of measures this year, including tighter scrutiny of overseas investments, AI startups, and cross-border technology transfers. Authorities have also investigated several Chinese AI companies that relocated abroad to determine whether they violated export control regulations.

While it remains unclear how China might restrict overseas access to its advanced AI models, legal experts have proposed a tiered regulatory framework. Under the proposal, basic open-source AI models would require only simple registration, more advanced models would undergo security reviews, and the most sensitive frontier AI systems could either be barred from public release or restricted for domestic use only.

(With inputs from Reuters )